VaBUS: Edge-Cloud Real-Time Video Analytics via Background Understanding and Subtraction

Abstract

Edge-cloud collaborative video analytics is transforming the way data is being handled, processed, and transmitted from the ever-growing number of surveillance cameras around the world. To avoid wasting limited bandwidth on unrelated content transmission, existing video analytics solutions usually perform temporal or spatial filtering to realize aggressive compression of irrelevant pixels. However, most of them work in a context-agnostic way while being oblivious to the circumstances where the video content is happening and the context-dependent characteristics under the hood. In this work, we propose VaBUS, a real-time video analytics system that leverages the rich contextual information of surveillance cameras to reduce bandwidth consumption for semantics compression. As a task-oriented communication system, VaBUS dynamically maintains the background image of the video on the edge with minimal system overhead and sends only highly confident Region of Interests (RoIs) to the cloud through adaptive weighting and encoding. With a lightweight experience-driven learning module, VaBUS is able to achieve high offline inference accuracy even when network congestion occurs. Experimental results show that VaBUS reduces bandwidth consumption by 25.0%-76.9% while achieving 90.7% accuracy for both the object detection and human keypoint detection tasks.Index Terms-edge-cloud collaborative computing, semantic compression, video analytics, task-oriented communication system I. INTRODUCTIONT HE recent advances of deep learning techniques boost the performance of many computer vision applications, such as object detection, semantic segmentation and keypoint detection. These advances facilitate the commercialization of video analytics applications including traffic control [1], video surveillance [2], and safety anomaly detection [3]. In order to make video analytics technologies ready to use, three main challenges need to be addressed, i.e., latency, bandwidth and accuracy. First, analytics applications require very low latency, Manuscript

[Retracted] Edge‐Enabled Heart Rate Estimation from Multisensor PPG Signals

“…Edge computing, as an emerging computing paradigm, provides an innovative opportunity for the synchronized computing of multisensor PPG signals in the proximity of the edge networks. Advancements in edge computing have the desired computing capacities for the rapid development of Internet of Tings (IoT) [1], Internet of Vehicles (IoV) [7,8], Satellite-Terrestrial Networks [9], reconfgurable wireless communications [10], and video streaming processing [11][12][13][14], especially in IoMT. Te explosive proliferation of IoMT generates massive amounts of time-series data, such as PPG signals.…”

Accuracy-Aware Cooperative Sensing and Computing for Connected Autonomous Vehicles

“…However, to support the data fusion and processing of selected object sensing data in our fine-grained cooperative sensing scheme, purely relying on a centralized edge server is not the most resource-efficient approach. As the object classification can be parallelized in per-object (or per region of interest) granularity, the computing subtasks for the classification of each object can be supported with distributed computing [19], [20], [21].…”